Embodiment
In the following description, at first use method of the present invention to determine rotor time constant, provided subsequently the description of estimated rotor resistance and threshold voltage about how.In further describing, in the situation that determine rotor time constant with other cache oblivious ground of induction machine, used the present invention, and provided at last about the description from the result calculating main inductance by method acquisition of the present invention.The description here also comprises the example be used to the control system of the frequency converter of implementing method of the present invention.
Rotor time constant
The present invention is based on following thinking: inactive when the power stage that makes frequency converter when motor is magnetized, that is, mains switch is controlled to be in the situation of off status, indirectly measure the decay counter voltage of induction machine.Before the power stage passivation that makes frequency converter or frequency converter, the magnetization of motor is performed, makes the magnetic flux of motor be in stable state.According to the present invention, provide the DC magnetization by frequency converter to induction machine, and the power semiconductor of frequency converter is controlled to be off status.
After following the power stage passivation closely, stator current commutates to the fly-wheel diode that is connected with the switch inverse parallel of controlling in the short period.Owing to there being voltage in motor terminal, the dc voltage that this voltage equals to use and having with respect to the opposite polarity of stator current (as long as diode loaded current), so current vector is rapidly reduced to zero.Current attenuation typically continues 200 to 300 μ s and rotor flux to zero and can greatly not descend during this period.
When electric current reached zero, the value of rotor flux was:
ψ
R=L
Mi
dc_magn=R
Rτ
ri
dc_magn (1)
I wherein
Dc_magnTo be fed to the magnetizing current of motor before making the power stage passivation.
After supposing passivation, at moment t=t
0Electric current has dropped to zero.After this moment, the rotor flux of motor is pressed timeconstantτ according to following index law
rDecay:
The decay of rotor flux has been inducted for the voltage of motor terminal, can use the dynamical equation of following motor to calculate this voltage:
U wherein
sdAnd i
sdStator voltage component and the stator current component on the direction of rotor flux and magnetizing current.
Because the derivative of electric current and electric current is zero, therefore according to (3):
Therefore the instantaneous voltage in electrode terminal is directly proportional to the value of rotor flux.Therefore, when rotor flux was decayed, stator voltage decayed according to same index law.By measuring two not in the same time terminal voltages of motor, will basically can determine rotor time constant.The voltage of equation (4) is low (typically less than 10V), to such an extent as in the situation that do not have expensive special device can not measure these voltages in frequency converter.
In the present invention, solved this problem by indirectly measuring the described moment.According to the present invention, all outputs of frequency converter are controlled as same current potential mutually, are used for providing zero-voltage vectors.So the power stage of frequency converter is activated and all export the positive or negative dc voltage that is controlled as mutually intermediate circuit, for generation of null vector.Due to the diode of power stage and transistorized threshold voltage less than induced voltage (4), the threshold voltage u that therefore equals to play a role on power semiconductor when the terminal voltage of motor
thThe time, stator current begins to flow.
u
sd=-u
th (5)
When using zero-voltage vectors, electric current is followed equation:
As shown in Figure 2, thus electric current by time constant σ L
s/ (R
s+ R
R) rise on the direction of original magnetizing current, this time constant is than rotor time constant τ
r=L
M/ R
RLittle a lot.After the specific time, electric current reaches its maximum i
Sdz_max, after this still decays due to rotor flux and force the derivative of the electric current in equation (6) finally to be born, so electric current begins to descend again.
In Fig. 2, electric current is at moment t=t
zHaving its maximum, is zero at the derivative of this moment electric current:
In the moment of electric current maximum, cancellation stray inductance item from equation (6).Therefore, at moment t
z, equation (6) can be written as:
Moment t
zThe value of rotor flux depend on as follows the maximum i of electric current
Sdz_maxAnd rotor time constant:
ψ
R(t
z)=(u
th+(R
s+R
R)i
sdz_max)τ
r (9)
If suppose at moment t
zBefore, can greatly the not slow down decay of rotor flux of the increase of the stator current during null vector, the index law of equation (2) should produce the value of the basis (9) about rotor flux.In other words:
Can solve rotor time constant from equation (10) as follows now:
By only (namely arriving t between the active period of null vector
z0) use index law and by calculating subsequently the change from the rotor flux of stator current, can consider because of period t
z0... t
zThe error that obtains in the valuation of the decay of the rotor flux that the increase of stator current during this time causes and equation (11):
At period t
z0... t
zComponent of voltage ψ in integration during this time
R/ τ
rValue can be approximately constant.Can be from moment t
zEquation (8) solve this value, this is the moment of electric current maximum constantly:
By the integral expression with equation (13) substitution equation (12), can as follows the rotor flux according to index law be extrapolated to t constantly
z:
Based on equation (14), should calculate rotor time constant, it is at moment t
zProvide the rotor flux with value identical with equation (9), that is:
As shown, can not solve more accurate rotor time constant from equation (15) in closing form.Yet, can use suitable alternative manner with its numerical value solve.Another kind may be to use the calculating valuation of rotor time constant on the right side of equation (15).For example can calculate this calculating valuation from the rated value of motor.Even rated value is slightly incorrect, calculates valuation in equation right side use and still produced than equation (11) valuation more accurately.
Calculate rotor time constant to need the resistance of stator and rotor be known from equation (11) or (15).The stator resistance that can measure with considerable accuracy during DC magnetization, but due to the Injection Signal that uses and frequency thereof, the rotor resistance valuation may have suitable error.Especially, error is by inaccurate the causing in kelvin effect and voltage measurement and/or voltage control.Due to the derivative of stator current at i
Sdz_maxNeighbouring approaching zero, so error can also be owing to determining t constantly
zInaccurate.
Different delay t when using null vector by utilization
z01And t
z02The pulse test that repeats to carry out above can reduce the sensitiveness from the valuation of equation (11) and (15).Illustrate this process in Fig. 3.Therefore, carry out twice different test, wherein at first pass through current i
Dc_maxMake motor magnetization and at moment t
0Make pulse deopiking.The test 1 at moment t
z01Utilize null vector, after this stator current is according to curve i
sd1Flow and at moment t
z1Reach its maximum i
Sdz_max1The test 2 at t
z01The moment afterwards is namely at moment t
z02Activate null vector, this after-current is according to curve i
sd2Flow and at moment t
z2Reach its maximum i
Sdz_max2
According to equation (9), moment t
z1And t
z2Rotor flux be:
If at period t
z1... t
z2Access failure null vector during this time, rotor flux will be decayed, and when pulse was passivated, the index law according to equation (10) made:
Can by carrying out identical approximate of situation with the valuation that utilizes equation (11), suppose at moment t now
z1And t
z2Can greatly the not slow down decay of rotor flux of the increase of the stator current during null vector before solves rotor time constant.In other words, satisfy equation group (16) according to the decay of the rotor flux (17) of index law, thus
By comparison equation (11) and (18), see the rotor time constant valuation of calculating by equation (18) than the valuation from equation (11), significantly the less error that depends on resistance.Along with threshold voltage u
thReduce, the sensitiveness of this error to the resistance valuation reduces.In addition, in two rotor fluxs that calculate according to equation group (16), null vector is approximately equalised on value to the relative effect of the decay of rotor flux in equation (18).This means, in fact compensated mentioned impact in the business of the logarithmic function of equation (18).Therefore, when testing to calculate rotor time constant with two independent null vectors, do not need the impact of method considering zero vector as about equation (15).
In addition, in the situation from the valuation of equation (18), can avoid from really regularly carving t in the valuation of equation (11) and (15)
zInaccurate.This is owing to the following fact: the rise time (t of electric current
z1-t
z01And t
z2-t
z02) be actually identical.Therefore, the time difference t in equation (18)
z2-t
z1Can be by time difference t
z02-t
z01Replace.Because time difference of the latter is accurately known valuation, therefore
Provided recently the valuation more accurate result from equation (18).
Therefore, according to the present invention, measure stator current during zero-voltage vectors, and determine the parameter of induction machine from measured stator current.
In the above-described embodiments, determined parameter is rotor time constant, and during null vector, has determined that also maximum current and the moment thereof are to be used for calculating.
According to embodiments of the invention, after the first test process, again make induction machine magnetization and utilize frequency converter shutdown and the null vector different period between connecting to repeat this test process.Reach time difference estimated rotor time constant between its peaked moment with electric current.Carry out therein in the preferred embodiments of the present invention of twice independent test, replace the time difference of maximum current between constantly with the time difference between the null vector switching time.As is understood, for the time difference between twice independent measurement, refer to the moment of point measurement from test separately, but not the time that disappears between test.
Rotor resistance and threshold voltage
In the embodiment that above explains of the present invention, estimated rotor time constant.In this was estimated, the parameter of using comprised rotor resistance and threshold voltage.Do not using these parameters that can provide in situation of the present invention for the estimated rotor time constant.Yet, the invention provides a kind of method than previously known and determine more accurately rotor resistance R
RWith threshold voltage u
thMethod.
Suppose to pass through current i
Dc_magnMake the motor magnetization until obtain stable state.As seen in Figure 4, remove pulse after magnetization, namely temporarily make the power stage passivation, and at moment t
z1Connect null vector, wherein electric current has dropped to value i
sd1Due to null vector, the decline of stator current at first stop and with after-current lentamente towards zero attenuation.In some cases, in Fig. 4, electric current increases temporarily, forms local maximum before it begins to descend.
By equation (3) as can be known, during null vector, the derivative of electric current is
Purpose is to find the extreme value i about stator current
sd0With corresponding connection moment t about null vector
z0, make the derivative of stator current of the right hand at moment t
z0Zero, namely
Subsequently, follow the connection of null vector closely constantly, following condition is set up:
Owing to working as electric current from i
Dc_magnDrop to i
sd0The time, in fact rotor flux does not change, therefore following approximate establishment:
ψ
R(t
z0)≈L
Midc_magn=τ
rR
Ri
dc_magn (22)
By with equation (22) substitution equation (21), obtained
R
Ri
dc_magn-u
th-(R
s+R
R)i
sd0=0 (23)
In other words, at moment t
z0The current limitation i of the zero derivative of generation current
sd0Value be
If electric current drops to this below limit before activating null vector, as in Fig. 4, be positive corresponding to the derivative of equation (21), t=t wherein
z1If in Fig. 5, again activate zero-voltage vectors, namely activate too early zero-voltage vectors, the electric current at switching time place is i
sd2>i
sd0And the derivative of electric current is born.
If at moment t
z0Connect zero-voltage vectors, wherein electric current drops to the value that equation (24) provides exactly, the derivative of the stator current of switching time be zero and current waveform consistent with Fig. 6.
Therefore purpose is to find the current limitation i of the current waveform that produces Fig. 6
sd0In fact, in order to find correct current value, must carry out a plurality of testing currents.Between test, the null vector switching time can change.Alternatively, can when dropping to the limit that sets, stator current utilize null vector.This limit also can change, and is used for obtaining correct current value, and the derivative of this current value constantly is zero connecting.
Be different from after magnetization and make the power stage passivation, if the connection that can set more accurately zero-voltage vectors is constantly, the voltage vector that can also force power stage to produce to have the direction opposite with the voltage vector of use during the DC magnetization.
Can search for relatively rapidly about connecting the extreme value that produces the electric current of zero derivative constantly.Can begin immediately the motor magnetization for next test after the electric current derivative that calculates null vector starting point place.Due to not too much decay of rotor flux during null vector, therefore the magnetizing time between test is also very short.The typical stator current waveforms of the searching period of zero derivative has been shown in Fig. 7.
Alternatively, from providing two current responses of positive derivative and negative derivative, can approximate current limit i
sd0If these current responses are as shown in Figures 4 and 5, can be by for example as follows, with derivative pro rata to current limitation i
sd1And i
sd2Weighting is from current limitation i
sd1And i
sd2Calculate the approximation of current limitation:
When finding optimal current response and relevant current limitation i
sd0The time, can use the current limitation from equation (23) to calculate the rotor resistance valuation, it provides:
The calculating of rotor resistance needs threshold voltage u
thWith stator resistance R
sKnown.The problem of equation (26) is that the value of threshold voltage is not necessarily accurately known.In addition, the value of threshold voltage greatly affects the accuracy of the valuation that can pass through equation (26) acquisition.Yet, if threshold voltage is known, can directly use equation (26).
Typically, in frequency converter, threshold voltage is be set to the parameter of steady state value or estimate threshold voltage during the DC magnetization.If the estimation threshold voltage, to postpone to make the threshold voltage valuation be not too accurately to the commutation in power stage, and if the threshold voltage u that therefore obtains from the DC magnetization
thBe placed in equation (26), the valuation of rotor resistance is also not too accurately.
Yet, can use equation (23) acquisition about the value more accurately (perhaps being used for the cancellation threshold voltage) of threshold voltage.In the situation that equation (23) is set up therein, null vector is activated, and electric current is uniform and it flows via power component, and the threshold voltage of these power components should be known in above equation.
The thinking of this estimation is to utilize no DC magnetizing current carry out above-described a plurality of null vector test and search for different magnetizing current i before making the power stage passivation
Dc_magn, iCurrent limitation i separately
Sd0, i, these current limitations are at the zero derivative of the connection moment of null vector generation current.Like this, regardless of the value of the magnetizing current that uses, equation (23) is set up.
For example, by using two different magnetizing currents, obtained a pair of equation:
And by cancellation threshold voltage u
th, rotor resistance can be calculated as the function of stator resistance:
Can also solve threshold voltage to (27) from equation, it provides:
Correspondingly, by using a plurality of different DC magnetizing currents, obtained a prescription journey
By using linear regression, can calculating and the proportional regression coefficient of threshold voltage and resistance ratio
And in addition, rotor resistance valuation and threshold voltage valuation can be calculated as the function of stator resistance.
Measure owing to utilizing a plurality of measurement points to carry out, therefore basically more accurate than the valuation that obtains by equation (28) and (29) by the valuation of (32) acquisition.
Estimated threshold voltage and rotor resistance can be identified individually and these valuations need to not used in embodiment given above.
The rotor time constant of cache oblivious
In above-described embodiment of estimated rotor time constant, stator and rotor resistance and be assumed to be known (equation (11), (15), (18) and (19)).In addition, equation (11) and (15) need rotor resistance as independent parameter.When the valuation of the rotor resistance that obtains by the above embodiment of the present invention or by some other modes and threshold voltage was placed in equation mentioned above, stator resistance was still unknown parameter.Can easily measure the stator resistance as the ratio between voltage and current during the DC magnetization.Yet typically by the dc voltage of intermediate circuit and determined by the output switch combination, the stator resistance of therefore measuring by this mode has certain uncertainty due to the average output voltage of frequency converter.
Because commutation postpones and the uncertainty of the threshold voltage compensation of power stage, the output voltage of frequency converter may be slightly inaccurate.Therefore because the voltage that is fed to motor is little, should inaccurately has during DC magnetizes and seriously influenced.This can be regarded as the error of the stator resistance of identification during DC magnetization, and in addition, if calculate rotor time constant based on stator resistance, can also be regarded as the error of rotor time constant.
When the embodiment of combination estimated rotor time constant mentioned above, rotor resistance and threshold voltage, can be almost and any cache oblivious ground calculating rotor time constant.In this embodiment of the present invention, carry out twice independent null vector test.In the first test (Fig. 8, test 1), pass through current i
Dc_magn1Make the motor magnetization until obtain stable state.Subsequently at moment t
0Remove pulse and at moment t
z0Start null vector.Due to null vector, electric current begins to increase and at moment t
zReach local maximum i
Sdz_max
After the first test, carry out the second stage of this embodiment.In second stage (test 2), determine that in conjunction with being used for the embodiment of rotor time constant and threshold voltage carries out similar null vector test.In this case, use the DC current i
Dc_magn2Make the motor magnetization, after this at moment t
0Remove pulse (making the power stage passivation).When electric current drops to the local maximum i that obtains in test 1
Sdz_maxLevel the time, activate null vector.Test 2 purpose is, when making the motor magnetization by mentioned electric current and in the situation that electric current activates null vector when being in the level of mentioned local maximum, finding to make current response have the magnetizing current i of zero derivative
Dc_magn2
By using above-described method be used to finding zero derivative (Fig. 4 to 7), make i
sd0Be replaced by i
Sdz_max, can search for suitable magnetizing current.Yet, in this case, be different from current limitation i
sd0, magnetizing current is changed.For example, in the situation of Fig. 4, magnetizing current should reduce, and in the situation of Fig. 5, magnetizing current should increase.
When the magnetizing current of the zero derivative that produces stator current is found, according to equation (23)
R
Ri
dc_magn2=u
th+(R
s+R
R)i
sdz_max (33)
When equation (33) (is made i by substitution equation (11)
Dc_magnBe replaced by i
Dc_magn1) time, obtained the valuation about rotor time constant:
As can be seen, in the formula according to equation (34), it is known that the parameter of electric machine needs not be.Due to similar equation (11), equation (34) is not considered period t
z0... t
zThe increase of the stator current the during decay of rotor flux during this time, so this formula comprises little error.This error depends on the parameter of electric machine, and on this meaning, equation (34) the also incomplete and parameter of electric machine has nothing to do.
Introduced consideration period t in equation (15)
z0... t
zThe valuation more accurately of the increase of stator current during this time.Also can use same equation in conjunction with the present embodiment., obtained during by substitution equation (15) when equation (33):
Similar equation (34), equation (35) is without any need for the parameter of electric machine.By using equation (35), can obtain the valuation more accurately of rotor time constant.On the other hand, the problem relevant to equation (35) be, rotor time constant be exist with implicit form and for it is solved, need iteration.Yet, by in the logarithmic function with the front valuation substitution equation right-hand side that once calculates always, easily untied this equation by iterative algorithm.Because the right-hand side of equation is not highly to depend on the rotor time constant that uses in the independent variable of logarithmic function, so iteration very rapidly restrains.
In theory, if the L equivalent electric circuit represents the behavioral characteristics of motor exactly, valuation and other parameters of electric machine of equation (35) generation are fully irrelevant.In fact, at moment t
0Iron loss is sizable on the impact of electric current derivative, and therefore provides the magnetizing current i of zero derivative
Dc_magn2The magnetizing current that should obtain in the time of may being different from the calculation of parameter of using the L equivalent electric circuit.
For this reason, preferably, as pointing out in Fig. 9, carry out twice null vector test, make pulse deopiking before this in certain period, make when null vector is activated electric current always zero.These the test basically with Fig. 3 in test similar.Yet the thinking in this situation is to use as follows two different magnetizing current i
Dc_magn1And i
Dc_magn2: the maximum of the stator current during null vector equates on value.In other words, in Fig. 9, i
Sdz_max1=i
Sdz_max2=i
Sdz_maxDue in the situation of Fig. 9, t
z02>t
z01So in test 2, magnetizing current i
Dc_magn2Must be higher than magnetizing current i
Dc_magn1Thereby, compensation period t
z01... t
z02The decay of rotor flux during this time.
By the hypothesis rotor flux based on equation (12) from moment t
0To moment t
z1And t
z2According to same index law decay, obtained
Current impulse during null vector is identical, and therefore
Equation has identical value to the right-hand side of (36), and has obtained
If work as magnetizing current from i
Dc+_magn1Be increased to i
Dc_magn2The time main inductance unsaturation, it is divided out from equation (37), making rotor time constant is unique the unknown.
Can followingly solve rotor time constant from equation (37):
Accordingly, in a preferred embodiment, at first by magnetizing current i
Dc_magn1Make the motor magnetization.The magnetization after at moment t
0The switch of frequency converter is controlled to be not on-state.At moment t
z01Controlling frequency converter generation null vector and stator current begins to rise.At moment t
z1Stator current reaches its maximum i
Sdz_max1And begin decay.
Use different magnetizing current i
Dc_magn2Motor is magnetized again and the process of duplication similarity.Purpose is to find magnetizing current and the corresponding time constant when utilizing null vector, and its generation has the peaked stator current that obtains corresponding in the phase I of testing.
May must search for desired magnetizing current by the second stage of retest.Magnitude variations by making magnetizing current or the moment of utilizing null vector is changed can change the maximum of the stator current during null vector.Figure 9 illustrates the second magnetizing current i
Dc_magn2Higher than the first magnetizing current i
Dc_magn1Yet the operation of this embodiment does not need at first to use lower electric current.Unique requirement about electric current is that electric current has different values.
In case find the peaked moment of the stator current during the second magnetizing current and null vector, can use equation (38) to calculate rotor time constant, the peaked moment of wherein also using the value of the first magnetized magnetizing current and the stator current during the phase I.
Because the current impulse during null vector is identical, therefore following condition is set up:
t
z1-t
z01=t
z2-t
z02
Therefore in equation (38), can differ from t service time
z02-t
z01Replace time difference t
z2-t
z1, caused
Due to than the time difference between maximum current, can determine more exactly the time difference between the connection constantly of null vector, therefore provided than equation (38) result more accurately from the valuation of the rotor time constant of equation (39).
By inferring at moment t
0+ t
z02-t
z01Last test in rotor flux ψ
R2Must equal first the test at moment t
0Rotor flux ψ
R1, can also obtain equation (39) by more direct mode.This is owing to the following fact, equal the period (=t at these after constantly
z01) expire and connect null vector afterwards, and their generations have the current impulse of same value exactly.
Therefore, can infer that following relation sets up and directly cause equation (39)
Main inductance
Can calculate long-pending main inductance L as rotor time constant and rotor resistance from the valuation that obtains by embodiments of the invention
M, namely
L
M=R
Rτ
r (41)
Because rotor time constant and the rotor resistance estimated by method of the present invention are also accurately, the main inductance that therefore calculates by equation (41) is accurately.
Control system
Figure 10 shows the example of the control system that can be used for implementing method of the present invention.This control system is to use between the mode by the expected of being used for is controlled error-free running period of frequency converter of motor.This control system produces the control signal for the power stage of frequency converter.Control the modulation scheme of motor in this control signal of modulator calculating of frequency converter according to being used for.
In Figure 10, the modulator that control system comprises current controller, voltage controller and operates in the dq coordinate system.During the DC magnetization, current controller output voltage benchmark, this current controller basis is about magnetizing current i
Dc_magnBenchmark control the d component of electric current, and correspondingly the q component of electric current is controlled to be zero.By these control actions, force current vector to be on the direction of angle θ, this direction has provided the direction of magnetization and has therefore provided the direction of d axle.Voltage reference is transformed in the stator reference system and is fed to voltage controller and further is fed to the modulator of power ratio control level.Be connected to the induction machine generation current response of frequency converter, based on the direction of measured phase current with magnetization θ, this current response transformed in the dq coordinate system.Preferably, angle θ is selected as being on the direction of phase place of induction machine, but also can at random be selected.
Can use this control system in conjunction with above-described definite rotor time constant as follows.At moment t=t
0(seeing Fig. 2), identification module is by being set as parameter m od_ena the power stage passivation that vacation (False) makes frequency converter, this is off status and when the terminal voltage of motor is set as the voltage of inducting by the decay of rotor flux with the output switch components set, and phase current is set as zero.At moment t=t
z0, identification module activates power stage and by parameter m od_ena and zero_force are set as very (True), forces the use zero-voltage vectors.During zero-voltage vectors, identification module is determined current i according to equation (11)
sdMaximum (i
Sdz_max) and moment t
z, be used for determining rotor time constant.
Can similarly implement other embodiment of the present invention by the control system of Figure 10 by sequential control enable signal zero_force and mod_ena.Signal zero_force forces and utilizes null vector (moment t
z0, t
z01And t
z02) and signal mod_ena make power stage passivation (t constantly
0).In the example of Figure 10, arrange null vector by providing voltage reference to modulator.Alternatively, also can utilize null vector and not use illustrated voltage reference in Figure 10 by direct control output switch element.
Describe in detail in conjunction with different embodiment as mentioned, different embodiment of the present invention may need DC magnetization, power stage passivation and the null vector of different number of times to force application.Recognition logic module in Figure 10 is paid close attention to the various signals of controlling process of the present invention and the value of monitoring stator current and derivative thereof, is used for obtaining about i
Sdz_max, i
Sdz_max1, i
Sdz_max2, i
sd, t
z, t
z1And t
z2Value.These values are notified to parameter calculating module by signal, and this parameter calculating module is used the various equations that are used for calculating as described in detail above desired parameters.
Above-mentioned control system has pointed out to calculate all above-mentioned parameters.Yet some parameters do not need to estimate with method of the present invention, and some parameters can be estimated with other method.
Be apparent that for those skilled in the art, along with technological progress, concept of the present invention can realize by variety of way.The present invention and embodiment are not limited to above-described example, but can change within the scope of the claims.